Automated Machine Learning Strategies for Multi-Parameter Optimisation of a Caesium-Based Portable Zero-Field...

Area of Science:

• Physics
• Engineering
• Computer Science

Background:

• Machine learning (ML) offers efficient parameter optimization for complex systems.
• Traditional methods are impractical for high-dimensional parameter spaces.
• Optically pumped magnetometers (OPMs) require precise parameter tuning for optimal sensitivity.

Purpose of the Study:

• To apply automated machine learning strategies for optimizing a single-beam caesium (Cs) spin exchange relaxation free (SERF) optically pumped magnetometer (OPM).
• To enhance the sensitivity of SERF OPMs through efficient parameter control.

Main Methods:

• Utilized automated machine learning (ML) strategies for OPM parameter optimization.
• Employed direct noise floor measurements and indirect on-resonance demodulated gradient measurements to guide optimization.
• Focused on optimizing operational parameters of the SERF OPM.

Main Results:

• Achieved a significant increase in optimal OPM sensitivity, improving from 500 fT/Hz to below 109 fT/Hz.
• Demonstrated the viability of both direct and indirect measurement methods for sensitivity optimization.
• ML approach proved efficient in navigating complex parameter landscapes.

Conclusions:

• Automated ML strategies are highly effective for optimizing SERF OPM sensitivity.
• The developed ML approach provides a flexible and efficient tool for benchmarking OPM hardware advancements.
• This method can accelerate the development of next-generation OPM sensors.